This article was written, and first published on this site, in May 2002.
The term schema (plural: schemata, or sometimes schemas) is widely used in cognitive psychology and the cognitive sciences generally to designate "psychological constructs that are postulated to account for the molar forms of human generic knowledge" (Brewer, 1999). The vagueness of this definition is no accident (and no sort of failing on Brewer's part). In fact schema is used in such very different ways by different cognitive theorists that the term has become quite notorious for its ambiguity (Miller, Polson, & Kintsch, 1984 p. 6). However, a concept of schema plays a crucial theoretical role in my account of the Perceptual Activity Theory of mental imagery (Thomas, 1999, 2000). Schema, as I use it, refers to a data structure, implemented in the brain, that functions to govern perceptual exploration of the world so that appropriate perceptual tests are applied at appropriate times and places, and that is continuously modified or updated by the results returned by those tests so as to be able to govern perceptual exploration more efficiently in the future (Thomas, 1999 §2.3.1). Confusion may arise, however, for readers familiar with other theorists' usages, who may take the term as I use it to have connotations that I do not intend, and that may not always be consistent with my actual views. Some clarification thus seems in order.
Although the different usages of schema are seldom completely unrelated, and have common etymological roots, it can by no means be safely assumed that properties or functions explicitly or implicitly attributed to schemata by one theorist will necessarily apply to schemata as another theorist conceives of them. But although these differences of usage may sometimes reflect genuine theoretical disagreement, often (I rather suspect it is more often) the difference seems to be "merely verbal": schema may be being used in different senses by different theorists because they are concerned with quite different aspects of cognition, that have little in common except that their explanation seems to demand a "molar" approach, in some sense of that even vaguer jargon term. (In psychology, "molar" refers to processes at a relatively high level of psychological or neural organization, by contrast with "molecular" explanations, that focus on very low level processes, such as the behavior of individual neurons.)
For example, so far as I can tell, and despite our common concern with the imagination, the schema that plays a key role in my Perceptual Activity theory of imagery is not at all the same sort of thing as the image schemata described in the works of Mark Johnson, George Lakoff, and their followers (e.g. Johnson, 1987; Lakoff & Johnson, 1999; Turner, 1996). However (and despite the irritatingly overblown claims that some of these authors are sometimes wont to make for the significance, originality and facticity of their theories) I am quite sympathetic to many aspects of their ideas about cognition (at the very least, we are all in the embodied cognition camp), and I certainly do not see any obvious contradictions between their theory and mine. We are simply using "schema" in different senses.
My conception of schema is taken directly from Neisser (1976), for whom it plays essentially the same theoretical role. Indeed, I developed my version of Perceptual Activity theory largely through the attempt to understand the implications, and work out some of the details and ambiguities, of Neisser's very suggestive and insightful, but sometimes rather vague and loosely formulated, theoretical ideas. In hindsight, perhaps my exposition (Thomas, 1999) would have been clearer if I had coined a new bit of jargon to replace the overused "schema", but nothing suitably pithy springs to mind, and, in any case, I did not want to distance myself from Neisser, or conceal my very real debt to his work .
Neisser himself (1976) seems to have drawn his concept of schema principally from Bartlett (1932) (although he cites several others who use the concept in ways related to his own usage, particularly Piaget (1952) and Rumelhart (1975)). Bartlett, in turn, credits the work of neurologist Henry Head (e.g. Head & Holmes, 1911) as his own source for the schema concept. Inevitably, each time the concept has passed from one theoretical context to another, its meaning has been subtly (and maybe sometimes not so subtly) transformed. According to Bisiach, Luzzatti, & Perani (1979) (who themselves use schema in a sense related to, but subtly different from, Neisser), Head, in his turn, took the concept from the 19th century German physiologist Wernicke (of area fame), who may fairly safely be presumed to have learned it from Kant's Critique of Pure Reason (Kant, 1781/1929). Unfortunately, Kant's brief remarks about schemata are, notoriously, one of the most obscure (though crucial) parts of his whole, complex philosophical system (indeed, Kant himself acknowledged this, describing the "schematism" as "an art concealed in the depths of the human soul, whose real modes of activity nature is hardly likely ever to allow us to discover" (1781/1929, A141-B181). Thus, although Johnson (1987) claims his usage of schema is inspired directly by Kant, I do not think he would deny that he has transformed the Kantian concept radically, elaborating it in ways that Kant himself thought would not be possible, and might very well not have approved.
But Kant may well have been too pessimistic. So far as I can see, in the sense in which I use the term, there ought to be no deep, in-principle difficulties in producing a computational simulation of a schema and its operations (even though, like Kant, I understand the schema's primary role to be the synthesizing of perceptual information into a suitably conceptualizable form). To put the point another way, I believe it ought to be possible and quite satisfactory to model human schemata (in my sense) as modifiable computational data structures, and to write computer programs (to control the perceptual activity of robots) in which such data structures actually do function as schemata. My meaning of schema thus overlaps quite significantly with what the term means to a thoroughgoing computationalist such as Rumelhart (1975, 1980; Rumelhart et al., 1986).
By contrast, I am very much inclined to think that the image schemata of the Johnson-Lakoff school are such that it would be impossible in principle to implement them as computational data structures. No doubt some aspects of their structure and functioning could be modeled on a computer; but just as a computer model of a thunderstorm is not really a thunderstorm, and nobody really gets wet, I think Johnson and Lakoff would agree that a computer model of an image schema could not be an image schema, could not facilitate any actual metaphorical understanding of anything by the computer.
This is not intended as a criticism of their theory, however, for I am not, myself, a thoroughgoing computationalist: I do not believe that the workings of the human mind can be straightforwardly identified with the computational operations of the brain.1 Ex hypothesis, my schemata are not things of which we have direct conscious experience. They are "mental" entities only in the derivative sense that they are necessary to the proper explanation of certain truly mental things, consciously experienced things such as mental imagery. Although I hold that the schema plays a crucial and necessary role in the generation of conscious mental imagery, it can only actually do this in virtue of its role in governing the perceptual explorations of an embodied, sentient being (human, animal, or, conceivably, robotic) situated in a real physical environment. Computational theory may be appropriate for explaining what goes on in my brain, but that is only part (albeit a crucial part) of the explanation of my mentality.
It appears to me that Johnson-Lakoff image schemata, unlike my schemata, are intended to be understood as truly mental entities, on a par with mental imagery (although, perhaps, more introspectively elusive). Whether or not we normally have conscious experience of them, they do seem to be conceived as bearing intentionality, and as at least in principle knowable via introspection. (In this regard they are like ordinary mental images; and unlike (my) schemata . . . or Hebbian cell assemblies, or weight space vectors, or most other explanatory posits of cognitive science.) On my view, the conscious experience that constitutes imagery depends not only on having a suitable schema active in the brain, but also on the embodiment and environmental situatedness of the organism doing the experiencing. Lakoff and Johnson (1999) are quite clear that they also regard embodiment as crucial to cognition, and they clearly conceive the nature and existence of their image schemata as being dependent upon embodiment. In fact, it may be that image schemata are best understood as a (rather special) sub-species of imagery simpliciter: If I were the arbiter of terminology, I might have called them schematic images: an especially "unsaturated" form of imagery, produced by simulating only the very earliest and most generally applicable stages of the process of a perceptual exploration: only ordering a few preliminary perceptual tests (see Thomas, 1999), and thus only representing a very broad class of things, defined by a quite limited number of global features. (Like Runeson (1977) I suppose that individual perceptual instruments may be able to probe not only for local details, but also for quite global features of the objects, scenes, or events being explored.) Being schematic, such images are suitable to provide "metaphorical" scaffolding for a diverse range of specific thought processes (such "metaphoric" structuring of thought is the central concern of the Lakoff-Johnson school, and it is what their image schemata are supposed to explain). These schematic images would lack most of the rich sensuous detail that arises from simulating a wide range of perceptual tests, and that make "richer" imagery experiences more consciously salient (and thus more amenable to introspective report). But (if Lakoff, Johnson, et al. are right) what "schematic images" (or image schemata) lack in qualitative wealth and direct conscious impact, they may make up in the generality of their relevance and their (consequent) functional importance in thinking.
1.This is because I do not believe that it is possible for computational symbols,
however physically realized (in neural tissue, or silicon, or whatever) to be
truly mental: either in the sense of being contents of consciousness
or in the very closely connected sense (Searle, 1992) of bearing "original
intentionality". If (as may well be) our mentality does in some way depend
upon computational processes in our brains, the representations manipulated by
those processes do not bear intentionality (for detailed and persuasive
argument to this effect see Horst (1996)), and thus are not the mental
representations that we experience when we think, remember, or imagine. (return
Bartlett, F.C. (1932). Remembering. Cambridge: Cambridge University Press.
Bisiach, E., Luzzatti C., & Perani D. (1979). Unilateral Neglect, Representational Schema and Consciousness. Brain (102) 609-618.
Brewer, W.F. (1999). Schemata. In Robert A. Wilson and Frank Keil (Eds.), The MIT Encyclopedia of the Cognitive Sciences. Cambridge, MA: MIT Press.
Head, H. & Holmes, G. (1911). Sensory Disturbances from Cerebral Lesions. Brain (34) 102-254.
Horst, S. (1996). Symbols, Computation and Intentionality: A Critique of the Computational Theory of Mind. Berkeley, CA: University of California Press.
Johnson, M. (1987). The Body in the Mind: The Bodily Basis of Meaning, Imagination, and Reason. Chicago: University of Chicago Press.
Kant, I. (1929). Critique of Pure Reason. (N. K. Smith, Trans. & Ed.) London: Macmillan. (German first edition of original work published 1787).
Lakoff, G. & Johnson, M. (1999). Philosophy in the Flesh: The Embodied Mind and its Challenge to Western Thought. New York: Basic Books.
Miller, J.R., Polson, P.G., & Kintsch, W. (1984). Problems of Methodology in Cognitive Science. In W. Kintsch, J.R. Miller, & PG. Polson (Eds.), Method and Tactics in Cognitive Science (pp. 1-18). Hillsdale, NJ: Erlbaum.
Neisser, U. (1976). Cognition and Reality. San Francisco: Freeman.
Piaget, J. (1952). The Origins of Intelligence in Children. New York: International Universities Press.
Rumelhart, D.E. (1975). Notes on a Schema for Stories. In D.G. Bobrow & A. Collins (Eds.), Representation and Understanding: Studies in Cognitive Science (pp. 185-210). New York: Academic Press.
Rumelhart, D.E. (1980). Schemata: The Building Blocks of Cognition. In R.J. Spiro, B.C. Bruce, & W.F. Brewer (Eds.), Theoretical Issues in Reading Comprehension (pp. 38-58). Hillsdale, NJ: Erlbaum.
Rumelhart, D.E., Smolensky, P., McClelland, J.L., & Hinton, G.E. (1986). Schemata and Sequential Thought Processes in PDP Models. In D. E. Rumelhart & J. L. McClelland (Eds.), Parallel Distributed Processing: Explorations in the Microstructure of Cognition: Vol. 2 (pp. 7-57). Cambridge, MA: MIT Press.
Runeson, S. (1977). On the possibility of "smart" perceptual mechanisms. Scandinavian Journal of Psychology (18) 172-179.
Searle, J.R. (1992). The Rediscovery of the Mind. Cambridge MA: MIT Press.
Thomas, N.J.T. (1999). Are Theories of Imagery Theories of Imagination? An Active Perception Approach to Conscious Mental Content. Cognitive Science (23) 207-245.
Thomas, N.J.T. (2000). A Non-Symbolic Theory of Conscious Content: Imagery and Activity. Presented at Tucson 2000 (Toward a Science of Consciousness) Conference, Tucson AZ, April 14th 2000.
Turner, M. (1996). The Literary Mind: The Origins of Thought and Language. New York: Oxford University Press.
Return to Home Page:
Imagination, Mental Imagery, Consciousness, Cognition:
Science, Philosophy & History.